JP3555168B2 - Video decoding device - Google Patents

Description

[0001]
[Industrial applications]
The present invention relates to a moving picture coding method and a coding apparatus, and a decoding method and a decoding apparatus that divide a screen into a plurality of regions and perform coding / decoding for each region.
[0002]
[Prior art]
In recent years, H.A. H.261, MPEG and other high-efficiency encoding methods for moving image signals have been standardized, and these methods are widely used. However, when such a method is applied to a moving image having a large number of pixels, such as a high-definition television, it is difficult for one encoder / decoder to process the entire screen in real time. Therefore, a method has been proposed in which a screen is divided into a plurality of regions and encoding / decoding is performed for each region.
[0003]
In this encoding method, as a method of combining the generated code strings into one, a method of simply connecting the code strings of each area, a method of inserting a marker code at the beginning of the code string of each area and connecting them, and the like. Proposed.
[0004]
[Problems to be solved by the invention]
However, the above-described conventional method has a problem that a code string cannot be decoded when the encoding method and the decoding method use different division methods or when the number of divisions differs. In addition, even though the code sequence for each divided screen satisfies the standard syntax, the entire code sequence combining the code sequences for each divided screen does not satisfy the standard syntax, and there is a problem in compatibility. Occurs.
[0005]
The present invention solves the above-mentioned conventional problems. Even when a screen is divided into a plurality of regions and coding is performed, a code sequence having the same syntax as that when coding without dividing the screen is performed. A method and apparatus for encoding a moving image for generating a video signal, and a code string having the same syntax as a code string generated by coding without dividing a screen, and dividing the code string into a plurality of decoders It is an object of the present invention to provide a moving picture decoding method and a decoding apparatus for decoding a full-screen moving picture.
[0006]
[Means for Solving the Problems]
In order to achieve this object, the moving image encoding method of the present invention divides a screen into a plurality of regions, and in a region including the upper left end of the screen, adds a screen header at the beginning of the code string of the divided screen. In addition, in an area other than the area including the upper left end of the screen, a header for the screen is not added to the leading part of the code string of the divided screen. And generating a code string for the entire screen by connecting the separated code strings to boundaries between screen divisions and between the screens in the coding order without dividing the screen.
[0007]
Further, the moving picture encoding apparatus of the present invention is a screen divider that divides a screen of an input image into a plurality of regions, and encodes a time series of the divided screen obtained by the screen divider, and performs screen division boundaries and An encoder group that outputs a control signal between screens, a memory that stores a code string output by the encoder, and an output switch that switches the memory that outputs the code string generated by the encoder. A code string synthesizer that switches and outputs the output group of the memory group, and a control signal output from the encoder separates a code string output from the encoder between screen division boundaries and between screens. Controller for controlling the output switching device as described above, and controlling the code string synthesizer so as to connect the outputs of the memory group to a boundary between screen divisions and between the screens in the order of generating a full screen code string. Structure with The it has.
[0008]
Further, the moving picture decoding method of the present invention, in which a code string of a moving picture is input, and based on a predetermined screen dividing method, the header information of a screen layer or higher is determined by the position information of the code string header. Treat as a code string of the area including the upper left end of the screen, generate a divided code string by dividing the code string to correspond to each area of the divided screen, decode the divided screen for each divided code string, It has a configuration in which the obtained divided screen groups are combined to reproduce a moving image of the entire screen.
[0009]
Further, the moving picture decoding apparatus of the present invention decodes a header of a screen layer or higher in an input code string, further searches a header of a layer lower than the screen, obtains position information in the screen at the time of detection, and obtains a header. A header detector that outputs information, a code sequence divider that divides the input code sequence into a plurality, a delay device that delays the input of the code sequence divider, and a screen division method that is determined in advance, According to the position information of the header obtained by the header detector, the headers on the screen hierarchy and above are treated as a code string of the area including the upper left corner of the screen, and input when the input code string belongs to a different area from the immediately preceding code string. A division controller that switches the output destination of the code sequence and controls the output switch so as to output a divided code sequence corresponding to each area of the divided screen; and a header of a screen hierarchy or higher decoded by the header detector. Using information from A decoder group for decoding the divided code string divided by the division controller, and a screen combiner for combining the divided screen group decoded by the decoder group into an entire screen. Have.
[0010]
[Action]
With this configuration, even when a screen of a moving image is divided into a plurality of regions and a moving image is encoded for each region, a code sequence having the same syntax as a code sequence when encoding without dividing the screen is used. Can be generated. Also, when decoding a code sequence of a moving image, even if the input code sequence is a code sequence obtained by encoding without dividing the screen, the input code sequence is divided and each divided code sequence is decoded, and the moving image of the entire screen is decoded. An image can be obtained. For this reason, the syntax of the code string does not depend on the number of screen divisions and the method of dividing the screen. Even when the number of screen divisions and the method of dividing the screen differ between the encoding device and the decoding device, decoding can be performed correctly. . In addition, if the code string generated in each divided screen at the time of encoding satisfies the standard syntax, the entire code string combining the code strings of each divided screen also satisfies the standard syntax, and compatibility is improved. Can be kept.
[0011]
【Example】
Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the present embodiment, H.264 is used as the encoding method. Considering a method of performing encoding using a set of pixels of about 16 × 16 (hereinafter, referred to as a macroblock) as a unit, such as H.261 or MPEG, a screen unit is a frame and a screen is divided into three parts as an example. .
[0012]
First, a first embodiment of a moving picture coding method according to the present invention will be described with reference to FIG. FIG. 1 is a schematic pattern diagram showing a method of encoding frames 1 and 14 in a moving image. Here, it is assumed that the frame 1 and the frame 14 are continuous frames in the order of the code string.
[0013]
First, each frame is vertically divided into three. With this operation, frame 1 is divided into divided frames 2, 3, and 4, and frame 14 is divided into divided frames 15, 16, and 17. Encoding is performed for each divided area. Each divided frame is divided into a set of macroblocks that are continuous in the horizontal direction (hereinafter, referred to as a slice) as a unit when a code string is generated. Here, it is assumed that the length of the slice is the same as the length of the divided frame in the horizontal direction. With this operation, the divided frame 2 is divided into slices 5 and 8 in order from the top, and the final slice is slice 11. The divided frames 3, 4, 15, 16, and 17 are similarly divided into slices.
[0014]
The divided frame 2 includes the macroblock at the upper left corner of the frame, that is, the top macroblock of the frame 1. Therefore, when a header is added, such as when the frame 1 is the first frame of a code string of a moving image or the first frame of a set of frames, a header corresponding to those cases is first generated. . Then, regardless of whether the header is generated, the header 22 of the frame 1 is generated. Subsequently, the slice 5 is encoded into a code string 5a, and the slice 8 is encoded into a code string 8a. However, it is assumed that the code string of each slice includes the header of the slice. Thereafter, slice encoding is performed in order, and finally, the slice 11 is encoded to become a code string 11a.
[0015]
Since the divided frame 3 does not include the macroblock at the upper left corner of the frame, no frame header is generated. Therefore, first, the slice 6 is encoded into a code string 6a, and subsequently, the slice 9 is encoded into a code string 9a. Thereafter, slice encoding is performed in order, and finally, the slice 12 is encoded to become a code string 12a. Similarly to the divided frame 3, the divided frame 4 does not include the macroblock at the upper left corner of the frame, so that no frame header is generated. Therefore, the code strings 7a and 10a and the code string are generated in order, and finally the code string 13a is generated.
[0016]
Next, encoding is performed on the divided frames 15, 16, and 17, similarly to the divided frames 2, 3, and 4. At this time, since the divided frame 15 includes the macroblock at the upper left corner of the frame, the header 22 of the frame 14 is added immediately before the code string 18a of the leading slice 18. Since the divided frame 15 is the same area as the divided frame 2, the code sequence of the divided frame 15 is generated following the code sequence of the divided frame 2. The header of the frame is not added immediately before the code string 19a of the first slice 19 of the divided frame 16 and the code string 20a of the first slice 20 of the divided frame 17.
[0017]
By the above operation, a code string of each divided area is generated as shown in FIG. The code sequence in the left region of the frame is a code sequence when a moving image of the divided frame is encoded. The code strings in the other areas are obtained by removing the headers of the layers above the frame from the code string when the moving image of the divided frame is encoded.
[0018]
Further, these code strings are connected for each slice between boundaries of frame division and between frames. That is, in frame 1, the code strings 5a, 6a, 7a, and 8a corresponding to each slice are connected in the order of slices 5, 6, 7, and 8 from the top. Then, after the code sequence 13a of the last slice 13 of the frame 1, the code sequence 14a of the first slice 14 of the frame 14 is connected. Thus, the code sequence shown in FIG. 1C can be generated. The order of the code string is the same as the order when the frame 1 and the frame 14 are encoded without being divided.
[0019]
Next, a second embodiment of the moving picture coding method according to the present invention will be described with reference to FIG. FIG. 2 is a schematic pattern diagram showing a method of encoding the frames 31 and 45 in a moving image.
[0020]
First, each frame is horizontally divided into three. With this operation, the frame 31 is divided into the divided frames 32, 33, and 34, and the frame 45 is divided into the divided frames 46, 47, and. Encoding is performed for each divided area. At this time, each divided frame is divided into slices as a unit for generating a code string. With this operation, the divided frame 32 is sequentially divided into slices 35, 36, 37, and 38. The divided frames 33, 34, 46, 47 and 48 are similarly divided into slices.
[0021]
The divided frame 32 includes the macroblock at the upper left corner of the frame 31. Therefore, when a header is added, such as when the frame 31 is the first frame of a code string of a moving image or the first frame of a set of frames, a header corresponding to those cases is first generated. . Then, regardless of whether the header is generated, the header 58 of the frame 31 is generated. Subsequently, the slices 35, 36, 37, and 38 are sequentially encoded, and become code strings 35a, 36a, 37a, and 38a, respectively. However, it is assumed that the code string of each slice includes the header of the slice.
[0022]
Since the divided frame 33 does not include the macroblock at the upper left corner of the frame, no frame header is generated. Therefore, first, the slice 39 is encoded to be a code string 39a, and then the slices 40 and 41 are encoded to be code strings 40a and 41a, respectively. Since the divided frame 34 does not include the macroblock at the upper left corner of the frame, similarly to the divided frame 33, no frame header is generated. Therefore, the code strings 42a, 43a, and 44a are sequentially generated.
[0023]
Next, encoding is performed on the divided frames 46, 47 and 48 of the frame 45 in the same manner as the divided frames 32, 33 and 34. At this time, since the divided frame 45 includes the macroblock at the upper left end of the frame, the header 59 of the frame 45 is added immediately before the code string 49a of the leading slice 49. The header of the frame is not added immediately before the code string 52a of the first slice 52 of the divided frame 47 and the code string 54a of the first slice 54 of the divided frame 48.
[0024]
By the above operation, the code string of each divided area is generated as shown in FIG. The code sequence in the region above the frame is a code sequence when a moving image of the divided frame is encoded. The code strings in the other areas are obtained by removing the headers of the layers above the frame from the code string when the moving image of the divided frame is encoded.
[0025]
Further, these code strings are connected for each slice between boundaries of frame division and between frames. That is, in the frame 31, the code string 39a of the slice 39 is connected after the code string 38a of the slice 38, and the code string 42a of the slice 42 is connected after the code string 41a of the slice 41. After the code sequence 44a of the slice 44, the code sequence 49a of the first slice 49 of the frame 45 is concatenated. As a result, the code sequence shown in FIG. 2C can be generated. The order of the code string is the same as the order when the frame 31 and the frame 45 are encoded without being divided.
[0026]
As described above, when a frame is divided into a plurality of regions, a code string separated between frames and boundaries between frames is generated for each divided frame, and a code string is generated for a divided frame including the upper left end of the frame. When the header of the frame is added to the beginning of the frame and the code strings separated in the order of encoding without dividing the frame are connected at the boundary of frame division and between frames, the moving images of all frames are syntactically It can be seen that a code string that can be regarded as encoded can be generated.
[0027]
Next, an embodiment of a moving picture coding apparatus according to the present invention will be described with reference to FIGS. FIG. 3 shows a screen divider 61, encoders 62, 63, 64, output switches 65, 66, 67, memories 68, 69, 70, 71, 72, 73, a code string synthesizer 74, and a division controller 75. 1 is a block diagram of a moving picture encoding device composed of
[0028]
First, the input image 76 is input to the screen divider 61. The screen divider 61 determines a frame division method based on the screen division control signal 84 output by the division controller 75, divides an input frame, and outputs divided image signals 77, 78, and 79. Here, it is assumed that the frame is vertically divided into three as shown in FIG. 1A, and the divided image signals are assumed to be divided image signals 77, 78 and 79 in order from the left region of the frame. That is, the divided image signal 77 includes the divided frames 2 and 15, the divided image signal 78 includes the divided frames 3 and 16, and the divided image signal 79 includes the divided frames 4 and 17. The divided image signals 77, 78, 79 are input to the encoders 62, 63, 64, respectively. The encoders 62, 63, 64 receive the size of the divided frames and the encoding conditions from the division controller 75 as the encoding control signal 85. Each encoder encodes the divided image signal. The encoder 62 encodes an area including the macroblock at the upper left corner of the frame, and thus adds a frame header to each head of the frame. If the encoding target frame is the first frame of a moving image or the first frame of a set of frames, a header is added before the first frame. For example, when encoding the divided frame 2, the header 21 of the frame 1 is generated.
[0029]
In this way, code strings 80, 81, and 82 for the divided image signals are generated. The configuration of the code sequence is as shown in FIG. 1B. The code sequence 80 is the code sequence in the left region of the frame, the code sequence 81 is the code sequence in the central region of the frame, and the code sequence 82 is the right sequence in the frame. Corresponding to the code string in the region of. The output memories of these code strings are switched by an output switch. This will be described with reference to a code string 80 generated by the encoder 62 as an example. The encoder 62 outputs the code string separation signal 86 to the division controller 75 after outputting the boundary of frame division or the last code string of the divided frame. Upon receiving the code string separation signal 86, the division controller 75 sends an output switching control signal 87 to the output switch 65 corresponding to the encoder 62. The output switching control signal 87 causes the output switch 65 to switch the output destination memory of the code string 80. That is, if the code string is output to the memory 68 before receiving the output switching control signal 87, the code string is output to the memory 69 after receiving the output switching signal. For example, if the code string 5a of slice 5 has been output to the memory 68, the code string 8a of slice 8 will be output to the memory 69. In this manner, the code string generated by the encoder 62 is input to the memory after being separated into frame division boundaries and between frames. By the same operation, the code sequence 81 generated by the encoder 63 is stored in the memories 70 and 71 by the output switch 66, and the code sequence 82 generated by the encoder 64 is stored in the memory 72 by the output switch 67. 73 is input.
[0030]
The code strings input to the memories 68, 69, 70, 71, 72, 73 in this manner are combined into one code string 83 by the code string synthesizer 74 and output. The code string synthesizer 74 connects the code strings by a code string synthesis control signal 88. The code string synthesis control signal 88 is a signal for instructing to link the code strings in the memory in the order of the code strings when the frame is not divided according to the frame division method. For example, assuming that the code strings 5a, 8a, 6a, 9a, 7a, 10a of slices are stored in the memories 68 to 73, the code string synthesizer 74 stores the code strings of these slices in 5a, 6a, 7a. , 8a, 9a, 10a in that order from the memory. As a result, a code string is generated as shown in FIG.
[0031]
With the above operation, the moving picture encoding apparatus of the present invention divides a screen and performs encoding processing in parallel by a plurality of encoders, but also performs the same synchronizing as a code string when the screen is not divided. A code sequence of the tax can be generated.
[0032]
Next, a first embodiment of a first moving picture decoding method according to the present invention will be described with reference to FIG. FIG. 4 is a schematic pattern diagram illustrating a method of dividing a code string of a moving image into a plurality of parts and decoding the divided part. Here, a case is shown in which a code string for two frames in the input code string is decoded.
[0033]
FIG. 4A shows an input code string. The headers 91 and 104 are headers higher than the frame layer. The code string of the slice is divided into a header and a code string of image data. For example, the slice code sequence 94 includes a slice header 92 and image data code sequence 93. The same applies to the code strings 95 to 103 and 105 to 113 of the other slices. This input code string is divided into three so as to correspond to the divided frames divided in the horizontal direction. This is performed by decoding the position information on the screen included in the header of the slice. That is, the header of the slice is searched from the beginning of the input code string, and the position information of the slice is obtained when the header is detected. Then, it is determined which divided frame belongs to the position information, and if it belongs to a divided frame different from the previous slice, the code string is divided at the head of the header of the slice. If it belongs to the same divided frame as the immediately preceding slice, the code string is not divided. However, the headers 91 and 104 for the frame hierarchy and above are included in the code string in the area above the frame.
[0034]
Here, the input code string is divided so as to correspond to the divided frames 115 to 117 and 119 to 121 in FIG. First, by detecting the header 92 of the slice and decoding the position information, it is found that the code sequence 94 of the slice belongs to the area of the divided frame 115. Similarly, it can be seen that the code strings 95 to 97 of the slice also belong to the area of the divided frame 115 by decoding the position information of the header. Therefore, the code string is not divided for the code strings 94 to 97 of the slice. Next, by detecting the header of the code string 98 of the slice and decoding the position information, the code string 98 of the slice belongs to the area of the divided frame 116 and belongs to a different area from the immediately preceding area. I understand. Therefore, the code string is divided at the head of the header of the slice code string 98. Since the code strings 98 to 100 of the slice belong to the area of the divided frame 116, the code string is not divided between them. In a similar manner, the code string is divided at the head of the header of the code string 101 of the slice. Thereafter, the code string is divided at the header of the header 104 and the header of the slice code strings 108 and 110, and three divided code strings corresponding to each area are obtained as shown in FIG. 4B.
[0035]
Subsequently, decoding is performed for each divided code string. By decoding the code sequence in the upper region of the frame, the divided frames 115 and 119 decode the code sequence in the central region, and the divided frames 116 and 120 decode the code sequence in the lower region. As a result, divided frames 117 and 121 are obtained. The obtained divided frames 115, 116, and 117 are combined to obtain a frame 114, and the divided frames 119, 120, and 121 are combined to obtain a frame 118.
[0036]
Next, a second embodiment of the first moving picture decoding method of the present invention will be described with reference to FIG. The difference between the second decoding method and the first decoding method is that the input code string shown in FIG. 4A is divided into code strings corresponding to respective regions as shown in FIG. 4C. . This will be described below.
[0037]
First, the header of the input code string is searched, and when a header higher than the frame layer is detected, the header is distributed to the code strings of each area. This operation continues until a slice header is detected. As a result, all divided code strings include a header of a frame or higher. Thereafter, when the header of the slice is detected, the code sequence of the slice is divided according to the position information of the header of the slice in the same manner as in the first decoding method.
[0038]
Consider a case where the input code string is divided so as to correspond to the regions of the divided frames 115 to 117 and 119 to 121 shown in FIG. In the input code string of FIG. 4A, first, the header 91 is detected. Since the header 91 is a header of a layer of the frame or higher, it is distributed to a code string of each area. The method of dividing the slice code strings 94 to 103 is as described in the first embodiment. Next, when the header 104 is detected, the header 104 is distributed to a code string of each area because the header 104 is a header of a frame layer or higher. Thereafter, the code strings 105 to 113 of the slice are divided by the method described in the first embodiment. As a result, a code string for each area as shown in FIG. The method of synthesizing the divided frames obtained by decoding FIG. 4C to obtain a full-screen moving image is the same as the first decoding method.
[0039]
From the above, by the first moving image decoding method of the present invention, by dividing and decoding a code sequence of a moving image obtained by encoding without dividing a frame, and combining the obtained divided frames, It can be seen that the original moving image can be obtained.
[0040]
Next, a first embodiment of a first moving picture decoding apparatus according to the present invention will be described with reference to FIGS. FIG. 5 is a block diagram of a moving image decoding apparatus including a header detector 131, a division controller 132, a delay unit 133, a code string divider 134, decoders 135, 136, and 137, and a screen synthesizer 138. It is. Here, the frame is horizontally divided into three so as to correspond to the regions of the divided frames 115 to 117 in FIG. 4D, and the upper region of the frame is decoded by the decoder 135, and the center region is decoded by the decoder 136. However, consider the case where the decoder 137 decodes the lower region.
[0041]
The input code string 139 is input to the header detector 131 and the delay unit 133 at the same time. The header detector 131 decodes the header of the input code string 139 at a frame level or higher and searches for the header of the slice. When the header of the slice is detected, the position information of the header is decoded and output to the division controller 132 as a header information signal 148. When the slice detected by the header detector 131 belongs to a divided region different from the immediately preceding slice, based on a predetermined frame division method, the division controller 132 divides the code sequence division control signal 149 into a code sequence. Output to the container 134. Upon receiving the code string division control signal 149 from the division controller 132, the code string divider 134 switches the output destination of the input code string 140 that has passed through the delay unit. That is, if the code sequence divider 134 outputs the input code sequence 140 as the divided code sequence 141 to the decoder 135 at a certain time, the code sequence division control signal 149 is input from the division controller 132 here. , And switches the output destination of the input code string 140 to output the divided code string 142 to the decoder 136. When the output destination of the input code sequence 140 is switched, the input code sequence 139 is converted by the delay unit 133 so that the header of the slice detected by the header detector 131 is included in the divided code sequence after the output destination is switched. The input to the divider 134 is delayed.
[0042]
The above operation will be described for the case where the input code string in FIG. In this case, the header detector 131 first decodes the header 91. This header 91 is input to the decoder 135. Subsequently, the slice header is searched, and the slice header 92 is detected. By decoding the position information of the slice header 92, it can be seen that the code sequence 94 of the slice belongs to the area of the divided frame 115. Therefore, the code sequence divider 134 outputs the input code sequence 140 to the decoder 135 following the header 91. Similarly, it can be seen that the code strings 95 to 97 of the slice also belong to the area of the divided frame 115 by decoding the position information of the header. Therefore, the code sequence divider 134 does not divide the code sequence for the code sequences 94 to 97 of the slice, and outputs the code sequence to the decoder 135. Next, the header of the code string 98 of the slice is detected, and its position information is decoded. Thus, it can be seen that the code sequence 98 of the slice belongs to the region of the divided frame 116 and belongs to a region different from the immediately preceding region. Accordingly, the division controller 132 outputs a code sequence division control signal 149 to the code sequence divider 134. The code sequence divider 134 switches the output destination of the input code sequence 140 to the decoder 136 according to the code sequence division control signal 149. Since the input code string 139 is delayed by the delay unit 133, this switching is performed at the head of the header of the code string 98 of the slice. Since the code strings 98 to 100 of the slice belong to the area of the divided frame 116, the code string is not divided between them, and these code strings are output to the decoder 136. Then, in a similar manner, the code sequence divider 134 divides the code sequence at the head of the header of the slice code sequence 101 and switches the output destination of the input code sequence 140 to the decoder 137. Thereafter, the code string is divided at the header 104 and the header of the slice code strings 108 and 110, and three divided code strings corresponding to each area as shown in FIG. 4B are obtained.
[0043]
The divided code strings 141, 142, and 143 thus divided are decoded by the decoders 135, 136, and 137, respectively. At the time of decoding, header information of a frame layer or higher is required. Since this is decoded by a header detector, the information and the division method are received as a decoded information signal 150 from the division controller 132. The decoded divided image signals 144, 145, and 146 are combined into all frames by the screen combiner 138, and become an output image 147. At this time, the screen synthesizing unit 138 obtains the screen synthesizing method from the division controller 132 as the screen synthesizing control signal 151.
[0044]
Next, a second embodiment of the first moving picture decoding apparatus according to the present invention will be described with reference to FIGS. The second embodiment is different from the first embodiment in that the input code string shown in FIG. 4A is divided into code strings corresponding to respective regions as shown in FIG. Is decoded by each decoder. This operation will be described below.
[0045]
First, the header detector 131 searches for a header of the input code string 139. When the detected header is a header of the frame hierarchy or higher, the division controller 132 outputs the code sequence division control signal 149 so that the code sequence divider 134 distributes the header to all of the decoders 135, 136, and 137. Output. Subsequently, when the header of the slice is detected, the position information of the header is decoded and output to the division controller 132 as a header information signal 148. The method by which the division controller 132 divides the code sequence of the slice is the same as in the first embodiment. Upon detecting a header at or above the level of the next frame, the code sequence divider distributes the header again to all of the decoders 135, 136, and 137. When the header of the next slice is detected, the code string is divided based on the position information of the header. This method is the same as the method described in the first embodiment.
[0046]
When the input code string of FIG. 4A is input, the header detector 131 first detects the header 91. Since the header 91 is a header of the frame hierarchy or higher, the code sequence divider 134 outputs the header 91 to all of the decoders 135, 136, and 137. Next, upon detecting the header 92 of the code sequence 94 of the slice, the code sequence divider 134 outputs the input code sequence 140 to only the decoder 135. The method of dividing the code string of the slice belonging to the subsequent frame 114 is the same as in the first embodiment. When the header 104 is detected, the header 104 is a header of a frame layer or higher, so that the code sequence divider 134 outputs the header 104 to the decoders 135, 136, and 137. The method of dividing the code string of the subsequent slice is the same as in the first embodiment. By the above operation, the divided code string of FIG. 4C is obtained.
[0047]
When the divided code strings in FIG. 4C are decoded by the decoders 135, 136, and 137, each of the divided code strings 141, 142, and 143 has a header that is higher than the layer of the frame. The header is obtained by decoding the header. Therefore, there is no need to obtain the decoding information from the division controller 132 as in the first embodiment. The screen combining process of the divided frames obtained by the decoders 135, 136, and 137 by the screen combiner 138 is the same as in the first embodiment.
[0048]
As described above, according to the first moving picture decoding apparatus of the present invention, the input code string is divided based on the position information of the header of the input code string, and the divided code strings are decoded in parallel. It can be seen that the divided images can be combined to obtain a full-screen moving image.
[0049]
Next, an embodiment of a second moving picture decoding apparatus according to the present invention will be described with reference to FIGS. FIG. 6 shows a delay unit 161, a code string divider 162, a memory 163, an input switch 164, a decoder 165 as a first decoder, and decoders 166 and 167 as a second decoder. It is a block diagram of the decoding apparatus of the moving image comprised from the screen synthesizer 168 and the division | segmentation controller 169. Here, it is assumed that the code sequence shown in FIG. 4A is input as the input code sequence 170, the frame is divided into three in the horizontal direction, and the area above the frame is decoded by the decoder 165, and the center area is decoded by the decoder. Suppose that the decoder 167 decodes the region 166 below.
[0050]
The input code string 170 is input to the input switch 164 and the delay 161 at the same time. The input code string 170 is read into the input switch 164 and the same data is read into the delay 161 at the same time. That is, when the input switch 164 does not read the input code string 170, the input code string 170 is not read into the delay unit 161. The input switch 164 receives the input code string 170 and the divided code string 172 output from the code string divider 162 as inputs, and outputs one of them to the decoder 165. In the initial state, the input switch 164 outputs the input code string 170. While the input switch 164 is outputting the input code string 170, the decoder 165 decodes the header information of the frame of the input code string 170 or higher, and searches for the slice header. When the slice detected by the decoder 165 belongs to a divided area different from the immediately preceding slice based on a frame dividing method determined in advance, the division controller 169 sends the detected slice to the code sequence divider 162. This is output as a code string division control signal 179. Since the code sequence 94 of the first slice belongs to the region above the frame, the code sequence division control signal 179 is transmitted so that the code sequence divider 162 outputs the code sequence as a divided code sequence 172.
[0051]
Upon receiving the code string division control signal 179, the code string divider 162 switches the output destination of the input code string 171 through the delay unit 161. That is, when the code sequence divider 162 outputs the input code sequence 171 as the divided code sequence 172 to the memory 163 at a certain point in time, if the code sequence division control signal 179 is input from the division controller 169, the input The output destination of the code string 171 is switched and output to the decoder 166 as the divided code string 173. When switching the output destination of the input code sequence 171, the delay unit 161 converts the input code sequence 170 into a code sequence so that the slice header detected by the decoder 165 is included in the divided code sequence after the output destination is switched. The input to the divider 162 is delayed.
[0052]
The decoder 165 further continues to detect the slice header, and outputs the header information signal 182 to the division controller 169 when the division of the input code string for one frame is completed. Upon receiving the header information signal 182 from the decoder 165, the division controller 169 uses the input switch control signal 180 to input the input of the input switch 164 so as to output the code sequence stored in the memory to the decoder 165. Switch. Then, the decoder 165 decodes the code string stored in the memory. In addition, the decoders 166 and 167 can perform decoding while the decoder 165 detects the header of the slice. At the time of decoding, the header information of the frame layer or higher is required. Since this is decoded by the decoder 165, the information and the division method are received as the decoded information signal 181 from the division controller 169. . The decoded divided image signals 175, 176, and 177 correspond to the divided frames 115, 116, and 117, respectively, are combined into all frames by the screen combiner 168, and the frame 114 is output as the output image 178. At this time, the screen synthesizer 168 obtains a screen synthesis method from the division controller 169 as a screen synthesis control signal 183.
[0053]
When the decoding of one frame is completed, the division controller 169 switches the input of the input switch 164 to the input code string 170. Then, the decoder 165 decodes the header of the layer higher than the layer of the next frame and detects the header of the slice, divides the input code string, and continues the decoding processing.
[0054]
The above operation will be described for the case where the input code string of FIG. In the initial state, the input switch 164 selects the input code string 170, and the code string divider 162 outputs the code string to the memory 163. The decoder 165 first decodes the header 91, and then searches for a slice header. When the slice header 92 is detected, by decoding the position information of the slice header 92, it can be seen that the code sequence 94 of the slice belongs to the area of the divided frame 115. Therefore, the code sequence divider 162 outputs the input code sequence 170 to the memory 163 following the header 91. Similarly, it can be seen that the code strings 95 to 97 of the slice also belong to the area of the divided frame 115 by decoding the position information of the header. Therefore, the code sequence divider 162 does not divide the code sequence for the code sequences 94 to 97 of the slice, and outputs the code sequence to the memory 163. Next, the decoder 165 detects the header of the code string 98 of the slice, and decodes the position information. Thus, it can be seen that the code sequence 98 of the slice belongs to the region of the divided frame 116 and belongs to a region different from the immediately preceding region. Therefore, division controller 169 outputs code sequence division control signal 179 to code sequence divider 162. The code sequence divider 162 switches the output destination of the input code sequence 170 to the decoder 166 according to the code sequence division control signal 179. Since the input code string 170 is delayed by the delay unit 133, this switching is performed at the head of the header of the code string 98 of the slice. Since the code strings 98 to 100 of the slice belong to the area of the divided frame 116, the code string is not divided between them, and these code strings are output to the decoder 166. In a similar manner, the code sequence divider 162 divides the code sequence at the beginning of the header of the slice code sequence 101 and switches the output destination of the input code sequence 170 to the decoder 167. That is, after the code string divider 162 outputs the code strings 94 to 97 of the slice as the divided code string 172 and detects the header of the code string 98 of the slice, the code string divider 162 converts the input code string into the divided code string 173. Will be output.
[0055]
When the decoder 165 detects the header 104, it means that the entire code string of the frame 114 has been divided. Therefore, the division controller 169 selects the output of the memory 163 as the input of the input switch 164 according to the input switching control signal 180. At this time, the code strings 94 to 97 of the slice are stored in the memory 163, the code strings 98 to 100 of the slice are input to the decoder 166, and the code strings 101 to 103 of the slice are input to the decoder 167. It will be. Then, the decoders 165, 166, and 167 decode the divided code strings input thereto. The decoded divided frames 115, 116, and 117 are combined with the frame 114 by the screen combiner 168.
[0056]
When the decoding of one frame is completed, the input switch 164 switches the input to the input code string 170, and the decoder 165 again decodes the header of the frame layer or higher, and searches for the header of the slice. Perform column splitting. By this division, the code strings 105 to 107 of the slice corresponding to the area of the divided frame 119, and the code strings 108 and 109 of the slice corresponding to the area of the divided frame 120 become slices corresponding to the area of the divided frame 121. Are obtained as divided code strings. Then, by decoding the divided code string, divided frames 119, 120, and 121 are obtained, and these are combined to obtain a frame 118.
[0057]
As described above, the second moving picture decoding apparatus of the present invention can divide an input code string and perform decoding in parallel, combine decoded decoded pictures to obtain a full-screen moving picture. We can see that we can do it.
[0058]
Next, an embodiment of the second moving picture decoding method according to the present invention will be described with reference to FIGS. FIG. 7 is a schematic pattern diagram showing a divided code string in a method of dividing and decoding a code string of a moving image into a plurality. Here, it is assumed that the frame 118 is encoded using the frame 114 as a reference frame, and the frame 118 is not referred to by another frame.
[0059]
FIG. 4A is an input code string. This input code string is divided so as to correspond to the divided frame divided in the horizontal direction, as shown in FIG. The division of the code string is performed in consideration of a range that can be referred to at the time of decoding. For example, it is assumed that the reference range of the frame 118 with respect to the frame 114 is one macroblock vertically, horizontally, and horizontally. At this time, when the code sequence of the frame 114 is divided, the code sequence is divided such that the code sequence of the region obtained by expanding the range of the divided frame vertically by one macroblock is distributed in each divided code sequence. The divided frame 115 includes slices 94a, 95a, 96a, and 97a. Since the reference from the divided frame 119 to the divided frame 115 extends to the slices 98a and 99a, the divided code sequence of the divided frame 115 includes the code sequence of the slice. 98 and 99 are included. Similarly, the divided code sequence corresponding to the divided frame 116 includes the code sequences 96, 97, 101, and 102 of the slice, and the divided code sequence 117 corresponding to the divided frame 117 includes the code sequence 100 of the slice. . Since the frame 118 is not used for reference from another frame, the code string is divided in the same manner as in the first decoding device.
[0060]
FIG. 7A shows the code string thus divided. When the divided code string of each divided area is decoded, divided frames 191, 192, 193, 119, 120, and 121 are obtained as shown in FIG. The frame 114 is obtained by deleting the overlapping portions of the divided frames 191, 192, and 193 and combining the divided frames 115, 116, and 117. In addition, since the divided frames 119, 120, and 121 have no overlapping portion, the frames 118 can be obtained by directly synthesizing them.
[0061]
As described above, according to the second moving image decoding method of the present invention, it is possible to divide an input code string and perform decoding in parallel, and combine the decoded divided images to obtain a full-screen moving image. We can see that we can do it.
[0062]
Next, an embodiment of the third decoding device of the present invention will be described with reference to FIGS. FIG. 8 shows a moving image composed of a header detector 131, a division controller 132, a delay unit 133, a code sequence divider 134, decoders 135, 136, and 137, a screen synthesizer 138, and a header information analyzer 201. It is a block diagram of a decoding device. Here, it is assumed that the frame is divided into three parts in the horizontal direction, and the upper region of the frame is decoded by the decoder 135, the center region is decoded by the decoder 136, and the lower region is decoded by the decoder 137.
[0063]
The input code string 139 is input to the header detector 131 and the delay unit 133 at the same time. The header detector 131 decodes the header of the input code string 139 at a frame level or higher and searches for the header of the slice. That is, when the code string of FIG. 4A is input, first, the header 91 is decoded. The header detector 131 outputs the header information 148 to the division controller 132 and the header information analyzer 201. The header information analyzer 201 determines a range in which each frame is referred to from other frames based on the reference relation of the frames, and outputs the determined range to the division controller 132 as a motion compensation information signal 202.
[0064]
When the header detector 132 subsequently detects the header of the slice, it decodes the position information of the header and outputs it to the division controller 132 as a header information signal 148. The division controller 132 determines an input code string division method based on a predetermined frame division method and a range of motion compensation obtained from the header information analyzer 201. For example, it is assumed that the frame 118 is encoded with reference to the frame 114 by the motion compensation of one vertical, horizontal, and right macroblock, and the frame 118 is not referred to by another frame. In this case, the code sequence of the frame 114 is divided in consideration of the range of motion compensation. The reference to the divided frame 115 of the divided frame 119 extends to the slices 98a and 99a. The reference to the divided frame 116 of the divided frame 120 extends to the slices 96a, 97a, 101a, and 102a, and the reference to the divided frame 117 of the divided frame 121 extends to the slice 100a.
[0065]
The division controller 132 divides the input code sequence 139 by determining the range of motion compensation in this way and outputting the code sequence division control signal 149 to the code sequence divider 134. Since the code strings 94 and 95 of the slice are not used for reference from another area, they are output only to the decoder 135. The code strings 96 and 97 of the slices are output to the decoders 135 and 136 because they are referenced from the divided frame 120 which is another area. Since the code strings 98 and 99 of the slices of the divided frame 116 are used for reference from the divided frame 119 which is another area, they are output to the decoders 135 and 136. Similarly, the slice code strings 100 to 103 are also divided in consideration of motion compensation and input to each decoder. Further, the code sequence of the frame 118 is divided so as to correspond to the divided frames 119, 120, and 121 without considering the motion compensation. As a result, the code string input to each encoder is as shown in FIG.
[0066]
When the decoders 135, 136, and 137 decode the code sequence of FIG. 7A, the code sequence of the reference image required for decoding is input for each region. Decryption can be performed without accessing the memory. By the decoding, the decoder 135 obtains the divided frames 191 and 119, the decoder 136 obtains the divided frames 192 and 120, and the decoder 137 obtains the divided frames 193 and 121. The screen synthesizer 138 deletes and combines the overlapping portions of the divided frames according to the screen synthesis control signal 151. As a result, 115, 116, and 117 are synthesized from the divided frames 191, 192, and 193, and the frame 114 is synthesized, and the divided frames 119, 120, and 121 are synthesized to obtain the frame 118.
[0067]
As described above, the third moving image decoding apparatus of the present invention can divide an input code string and perform decoding in parallel, combine decoded decoded images to obtain a full-screen moving image. We can see that we can do it.
[0068]
Next, an embodiment of a fourth moving picture decoding apparatus according to the present invention will be described with reference to FIGS. FIG. 9 shows a delay unit 161, a code sequence divider 162, a memory 163, an input switch 164, a decoder 165 as a first decoder, and decoders 166 and 167 as a second decoder. FIG. 3 is a block diagram of a moving image decoding apparatus including a screen synthesizer 168, a division controller 169, and a header information analyzer 211. Here, it is assumed that the code sequence shown in FIG. 4A is input as the input code sequence 170, the frame is divided into three in the horizontal direction, and the area above the frame is decoded by the decoder 165, and the center area is decoded by the decoder. Suppose that the decoder 167 decodes the region 166 below.
[0069]
The input code string 170 is input to the input switch 164 and the delay 161 at the same time. The input switch 164 receives the input code sequence 170 and the divided code sequence 172 output from the code sequence divider 162 as inputs, and outputs one of them to the decoder 165. In the initial state, the input switch 164 outputs the input code string 170. While the input switch 164 is outputting the input code string 170, the decoder 165 decodes the header information of the frame of the input code string 170 or higher, and searches for the slice header. That is, first, the header 91 is decoded. The decoder 165 outputs the header information signal 182 to the division controller 169 and the header information analyzer 211. The header information analyzer 211 determines a range in which each frame is referred to from other frames based on the reference relation of the frames, and outputs the determined range to the division controller 169 as the motion compensation information signal 212.
[0070]
Subsequently, the decoder 165 searches the header of the slice. When the header 92 of the slice is detected, the position information of the header is decoded, and this is notified to the division controller 169 as a header information signal 182. The division controller 169 determines an input code string division method based on a predetermined frame division method and the range of motion compensation obtained from the header information analyzer 211. This dividing method is the same as the method described in the third embodiment of the present invention.
[0071]
The division controller 132 divides the input code sequence 139 by outputting a code sequence division control signal 149 to the code sequence divider 134 based on the determined division method. When the code string of the frame 114 is divided by the method described in the third embodiment of the decoding device of the present invention, the decoder 165 outputs the header information signal 182 to the division controller 169. Upon receiving the header information signal 182 from the decoder 165, the division controller 169 switches the input so that the code sequence stored in the memory is output to the decoder 165 by the input switching control signal 180. At this time, the code strings 94 to 99 of the slice are stored in the memory 163, the code strings 96 to 102 of the slice are input to the decoder 166, and the code strings 100 to 103 of the slice are input to the decoder 167. Will be. Then, the decoder 165 decodes the code string stored in the memory. In addition, the decoders 166 and 167 can perform decoding while the decoder 165 detects the header of the slice. At the time of decoding, the header information of the frame layer or higher is required. Since this is decoded by the decoder 165, the information and the division method are received as the decoded information signal 181 from the division controller 169. . Each of the decoders 165, 166, and 167 performs decoding by using only an image obtained by decoding the divided code string as a reference image.
[0072]
The decoded divided image signals 175, 176, and 177 correspond to the divided frames 191, 192, and 193, respectively, are combined into all the frames by the screen combiner 168, and the frame 114 is output as the output image 178. At this time, the screen synthesizer 168 receives the screen synthesis method from the division controller 169 as the screen synthesis control signal 183, and deletes the overlapping portion of the divided frames and synthesizes them. Therefore, the divided frames 191, 192, and 193 become the divided frames 115, 116, and 117, respectively, and are synthesized.
[0073]
When the decoding of one frame is completed, the division controller 169 switches the input of the input switch 164 to the input code string 170. Then, the decoder 165 decodes the header 104, detects the header of the slice, and divides the input code string. Since the frame 118 is not referred to from other screens, it is not necessary to consider the range of motion compensation when dividing the input code string. Therefore, the slice code strings 105 to 107 are input to the memory 163, the slice code strings 108 and 109 are input to the decoder 166, and the slice code strings 110 to 113 are input to the decoder 167 as the split code strings. You. When the division of the code string of the frame 118 is completed, the decoder 165 notifies the division controller 169 by the header information signal 182. The division controller 169 switches the input of the input switch 164 to the output of the memory. Then, the decoders 165, 166, and 167 decode the divided code string. Since the decoded divided frames 119, 120, and 121 have no overlapping portion, they are synthesized by the screen synthesizer 168 as they are, and the frame 118 is output as the output image 178.
[0074]
As described above, the fourth moving image decoding apparatus of the present invention can divide an input code string and perform decoding in parallel, combine decoded decoded images to obtain a full-screen moving image. We can see that we can do it.
[0075]
In the present embodiment, a case where a frame is used as a unit of a screen has been described, but this may be a field.
[0076]
In the embodiment described with reference to FIG. 1, the length of the slice does not need to be the same as the horizontal length of the divided screen, and may be a slice shorter than the horizontal length of the divided screen.
[0077]
The method of dividing the slice may be freely determined depending on each divided screen.
Further, in the present embodiment, the case where the screen is divided by the horizontal division and the vertical division has been described, but this may be a cross division.
[0078]
Further, in this embodiment, the case where the screen is divided into three is described, but this may be an arbitrary number of divisions.
[0079]
In this embodiment, H.264 is used as a basic encoding method. 261, the case of using a method of performing coding in units of macroblocks such as MPEG has been described, but this may be another coding method as long as the coding method includes the position information of the screen in the header.
[0080]
Also, in the second decoding method of the present invention, the headers of the screen hierarchy and higher are included in the divided code string of the area including the upper left end of the screen, but this is the same as the first decoding method of the present invention. As described in the second embodiment, it may be included in all divided code strings.
[0081]
Also, in the second, third, and fourth decoding devices of the present invention, the headers at or above the screen layer are input only to the decoder that decodes the region including the upper left corner of the screen. As described in the second embodiment of the first decoder of the present invention, the data may be input to all decoders.
[0082]
【The invention's effect】
As described above, according to the present invention, at the time of encoding, a screen is divided into a plurality of areas, and in an area including the upper left end of the screen, a screen header is added to the head of the code string of the divided screen, Generates a code string separated between screen division boundaries and screens for each screen, and concatenates the separated code strings between screen division boundaries and between screens in the order of encoding without dividing the screen. Generate a code sequence. Also, at the time of decoding, a code string of a moving image is input, and based on a predetermined screen division method, a header at a layer higher than the screen includes the upper left end of the screen according to position information of the header of the code string. Treat as a code string of an area, generate a divided code string by dividing the code string so as to correspond to each area of the divided screen, decode the divided screen for each of the divided code strings, and obtain the obtained divided screen group. Combine and play back the full-screen moving image. Therefore, it is possible to have the same syntax as the code string generated when encoding the code string without dividing the screen. At the time of decoding, even if the input code string is a code string obtained by encoding without dividing the screen, it can be divided and decoded. For this reason, the syntax of the code string does not depend on the number of screen divisions and the screen division method, and decoding can be performed correctly even when the number of screen divisions or the screen division method differs between the encoding device and the decoding device. .
[Brief description of the drawings]
FIG. 1 is a schematic pattern diagram of an embodiment of a moving image encoding method according to the present invention.
FIG. 2 is a schematic pattern diagram of an embodiment of a moving image encoding method according to the present invention.
FIG. 3 is a block diagram of an embodiment of a moving picture encoding apparatus according to the present invention.
FIG. 4 is a schematic pattern diagram of an embodiment of a first moving picture decoding method according to the present invention;
FIG. 5 is a block diagram of an embodiment of a first moving picture decoding apparatus according to the present invention;
FIG. 6 is a block diagram of an embodiment of a second moving picture decoding apparatus according to the present invention;
FIG. 7 is a schematic pattern diagram of an embodiment of a second moving picture decoding method according to the present invention;
FIG. 8 is a block diagram of an embodiment of a third moving picture decoding apparatus according to the present invention.
FIG. 9 is a block diagram of an embodiment of a fourth moving picture decoding apparatus according to the present invention;
[Explanation of symbols]
1,14 frames
2-4, 15-17 divided frame
5-13, 18-20 slices
5a to 13a, 18a to 20a Code string of slice
21, 22 header
31, 45 frames
32-34, 46-48 split frame
35-44, 49-57 slices
35a-44a, 49a-57a Code sequence of slice
58, 59 header
61 screen divider
62-64 encoder
65-67 output switch
68-73 memory
74 code string synthesizer
75 split controller
91, 104 header
92 slice header
93 Code string of image data
94-103, 105-113 Slice code string
94a-103a, 105a-113a slice
114, 118 frames
115-117, 119-121 Divided frame
131 header detector
132 split controller
133 delay unit
134 code string divider
135-137 Decoder
138 Screen Synthesizer
161 delay unit
162 code sequence divider
163 memory
164 input switch
165-167 Decoder
168 Screen Synthesizer
169 division controller
201, 211 header information analyzer

Claims (4)

Decoding the header of the entire moving image, the header of a frame set, or the header of a frame (hereinafter, referred to as a large header) in the input code string, and further detecting a slice header added to a slice constituting the frame; A header detector that outputs position information indicating which position in the frame the image data is ;
And code stream divider for dividing the input code string into a plurality of divided code string,
A delay unit for delaying an input of the code sequence divider;
The frame is divided into a plurality of regions in the horizontal direction, the large header is output together with a divided code string of a region including an upper left end of the frame , and a code of a slice including the position information obtained by the header detector is output. When a sequence (hereinafter, referred to as a code sequence of the current slice) belongs to a different region from the code sequence of the slice decoded immediately before the code sequence of the current slice, the output of the input code sequence is output. A division controller for controlling the code sequence divider so as to output a divided code sequence corresponding to an area to which the code sequence of the decoding target slice belongs ;
A decoder group that decodes the divided code string output from the code string divider using the information of the large header decoded by the header detector,
A moving image decoding apparatus, comprising: a screen synthesizer for synthesizing a divided screen group decoded by the decoder group into a frame. For the reference frame used as a reference frame in decoding of another frame , the division controller adds a code sequence of a slice referenced from the other frame to a code sequence of each region of the frame . Output the divided code string,
For a frame that is not referenced from the other frames , the code string divider is controlled to output a divided code string by dividing a code string so as to correspond to each region of the frame ,
The group of decoders, decoding with reference to only the image obtained by decoding the divided code string output by the code string divider,
2. The decoding apparatus according to claim 1, wherein the screen synthesizer deletes a portion corresponding to the code string of the added slice from an image decoded by the group of decoders and synthesizes the image into a frame . apparatus. 3. The moving picture decoding apparatus according to claim 1, wherein one of the decoder groups operates as the header detector. Wherein in the code string in the region other than the region including the upper left end of the screen included a large header, the decoding apparatus of the moving image according to claim 3, wherein said decoder group decoding of the large header performed.

Images